This randomized comparative effectiveness trial examines whether active manipulation of a key component of semantic feature analysis (SFA) treatment for word-finding difficulty in aphasia improves outcomes. The key component in question is the number of semantic features that persons with aphasia are asked to generate on each treatment trial. Study participants (n=40) will be recruited and randomized to receive either a many- features version of SFA or a few-features version. In the many-features condition, participants will be asked to generate 11 semantic features for each word practiced. Participants assigned to the few-features condition will be asked to generate 5 features for each word practiced. The total treatment time will be equated in the two conditions. Because each trial will take less time in the few-features condition, participants in this group will cycle through the lists of treated items more often, providing them with more opportunities to practice the phonological form of the targets, at the expense of more elaborated feature generation practice. Correspondingly, the many-features group will receive more practice generating semantic features, at the expense of few opportunities to practice the target word forms. Study participants will be housed locally at the Pittsburgh site for five weeks during which they will receive 60 hours of SFA treatment with pre- and post-treatment assessment of their ability to name pictures of treated and untreated, semantically related nouns. Other secondary outcomes, including measures of connected speech and patient-reported communication ability will also be collected. In order to address unresolved questions about the underlying cognitive and neural mechanisms of SFA, participants will also receive concurrent pre- and post-treatment assessment of automatic word processing ability using eye-tracking methods and functional magnetic resonance imaging (fMRI). Participants will also be asked to return to Pittsburgh for one-month follow-up language, eye-tracking, and fMRI testing. The language testing results will be used to determine the appropriate balance of feature generation practice vs. word form practice to optimize SFA outcomes. The eye-tracking results will be used to infer whether SFA?s positive effects can be attributed to improved activation of lexical-semantic representations, improved ability to inhibit competing representations, or both. The fMRI results will be used to identify the brain networks and activation changes associated with changes in naming ability resulting from SFA. This study will provide theoretically and clinically relevant information about how aphasia treatment should be delivered and the neurocognitive mechanisms underlying its effects.